|Publication number||US4090238 A|
|Application number||US 05/729,348|
|Publication date||May 16, 1978|
|Filing date||Oct 4, 1976|
|Priority date||Oct 4, 1976|
|Also published as||DE2744531A1, DE2744531C2|
|Publication number||05729348, 729348, US 4090238 A, US 4090238A, US-A-4090238, US4090238 A, US4090238A|
|Inventors||Paul Michael Russo|
|Original Assignee||Rca Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (7), Referenced by (43), Classifications (9), Legal Events (2)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This invention relates to stored program digital computers having an interrupt facility.
In the stored program digital computer art, an interrupt is understood to mean a temporary suspension or break in the normal operation of executing sequential instructions so that the processor can service an external device or branch to a predetermined place in the instruction stream in response to an external signal. Peripheral devices providing data storage used in conjunction with the processor operate at speeds which are very slow in comparison to the speed of operation of the processor. Interrupts provide a means for permitting the processor to execute instructions while waiting to service the peripheral devices, i.e., to provide data to be written to the device, to accept data read from the device, or to terminate the device's operation. Another use of interrupts makes programs more versatile and general in application by permitting the operator to indicate which of several optional branches should be taken during the execution of the program depending on some external criteria. This is sometimes accomplished by switches, called break points, which are sensed by the machine using special instructions for that purpose.
There is usually only one interrupt input signal in a processor. When several devices are capable of interrupting, their interrupt request signals are ORed together to provide a single processor interrupt signal. Activation of the interrupt signal causes the processor to branch to a first subroutine that determines which device caused the interrupt signal and which service subroutine should next be executed. One method of identifying the interrupting device is polling where external flag lines are used to identify the device or where each device is interrogated in turn by a special instruction until an active device is found.
A priority interrupt system is used in systems having devices that must be serviced at a faster rate than others. For example, a high-speed magnetic tape reader must be serviced more often than a paper tape reader because the character rate is greater and allowable buffer time is shorter because of the higher data flow speed. The problem is one of determining not only which device requested an interrupt but also which of several interrupting devices has the highest priority and should therefore be serviced first.
A vectored interrupt is an interrupt providing a plurality of responses to the interrupt signal. Sometimes several interrupt lines are provided. In the case of a single interrupt line, a subroutine is used to scan or to poll the possible sources of interrupt and to determine from the results which subroutine or program branch is to be executed.
Prior art systems utilizing interrupts require substantial software support which increases the processor time required for vectored interrupts. If priority interrupts are also necessary, more complex software is required and the servicing of interrupts uses more processor time. In microprocessors, where the instructions are usually more basic than in larger machines, many instructions may be required to perform priority or vectored interrupt servicing, or both.
Some computer processing units include a direct memory access (DMA) facility for storing data from external sources directly into the memory or for retrieving data directly from the memory without disturbing the execution of the current program. This is usually accomplished by generating a special machine cycle, logically called a DMA cycle, which couples a register known as a pointer to the memory address bus and which stores data in the memory or retrieves data from the memory at the addressed location depending on whether the DMA request was for input or output. Since the program execution requires certain cycles, the occurrence of the DMA cycle merely suspends for one cycle time the operation of the normal program execution. This is sometimes referred to as cycle stealing.
An interruptable processor embodying the invention and having a DMA facility can be made to perform a vectored interrupt subroutine transfer with a minimum number of cycles. In operation, the service request signals from peripheral devices are applied to latches. When one of these latches is set, an interrupt request signal is generated. When the interrupt occurs, a direct memory access cycle is requested. During the DMA cycle, a priority determining circuit activates one of several signals representing the highest priority device requesting service and gates an associated predetermined address onto the data bus of the machine.
In the drawings:
FIG. 1 is a logic diagram of an embodimdent of the invention useful in a system requiring only one data word to specify a subroutine address;
FIG. 2 is a logic diagram of a modification of the circuit of FIG. 1 for use in a system where a subroutine address requires two data words; and
FIG. 3 is a block diagram of a modification of a portion of the system of FIG. 1.
The preferred embodiment of the invention accomplishes a transfer to a subroutine related to the highest priority interrupting device by utilizing the DMA facility of the associated processor in connection with the interrupt mechanism. To describe the detailed operation of the invention, it is assumed for purposes of explanation that the preferred embodiment is used in conjunction with a COSMAC type microprocessor such as the CDP1802 (RCA Corporation). The detailed operation of a COSMAC-type microprocessor can be found in the data sheets and application notes from the manufacturer and in U.S. Pat. No. 3,970,998, assigned to the same assignee as this application. The details necessary for an understanding of the invention are included in the explanation below.
The COSMAC microprocessor has several internal states depending on the type of cycle being executed. For instance, a FETCH cycle retrieves the next instruction to be executed from the associated memory. The EXECUTION cycle performs the instruction fetched during the previous fetch cycle. Most of the COSMAC instructions require only one execution cycle although there are a few which require two such cycles. Another internal state is an INTERRUPT cycle during which the contents of the current program counter and operand pointer are stored in a special register and predetermined registers in the register array are assigned as the new program counter (R1) and as the new operand pointer (R2). The interrupt cycle is entered in response to an interrupt request signal and, in effect, suspends the program being executed and branches to another program addressed by the contents of R1.
Another internal state is the DMA cycle which is instituted by a DMA request signal. There are two DMA request signals, a DMA-IN and DMA-OUT, depending on whether the memory is to be written into or read from. A special register (RO) is the DMA pointer. This is preset to the address to which or from which the data during a DMA-IN or a DMA-OUT cycle is to be accessed. The R0 pointer is automatically incremented after each DMA cycle.
During each cycle, two time pulses (TP) are supplied from the COSMAC microprocessor. One is a TPA pulse which occurs near the beginning of the cycle. The other is a TPB pulse which occurs near the end of the cycle. The time pulses and internal state of the microprocessor are made available on external connectors, the type of cycle being indicated by means of state code signals which have a unique combination for each internal state.
To utilize the circuit of the invention, the processor's DMA pointer (R0) is initially set to the address of the address portion of a branch instruction which is addressed by the R1 interrupt pointer. By following an interrupt request with a DMA-IN request, the address to which the initial branch instruction of the interrupt routine will transfer is the location specified by the word on the data bus during the DMA cycle. A short branch instruction requires only a single byte (eight-bit) address and transfers to a location on the same "page" as the branch instruction. (A page is a group of memory locations having the same high order address bits.) A long branch instruction requires a two-byte address and transfers to any location in the memory.
In the circuit of FIG. 1, the service request signals from several devices each set a particular flip-flop of a group of flip-flops or latches 11. The service request signals are gated to the corresponding latches by AND gates 12 which are primed by the output signal from an inverter 14 when no interrupt request (INT REQ) signal is present. The interrupt request signal is generated by an OR gate 15 which is the last in a group of cascaded OR gates such as the OR gates 16a and 16b. The effect of the cascaded OR gates is to produce an interrupt request at the output terminal of the OR gate 15 when at least one of the latches 11 is set.
It is assumed in FIG. 1 that Device 1 has the highest priority, Device 2 has the next lower priority, and so on, to Device n which has the lowest priority. If several interrupt request signals occur simultaneously, the highest priority will be controlling as will be seen from the following description. If Device 1 produces a service request signal while no interrupt request is generated, the set output signal from the associated latch will prime an AND gate 18a, activate an OR gate 16b, and inhibit a NOR gate 17a. (A NOR gate operates like an AND gate considering both inputs to be true when negative as indicated by the inversion circles at the input terminals.) The output signal from the OR gate 16b inhibits a NOR gate 17b and activates the OR gate 16a. In a similar fashion, all of the NOR gates 17a-17c are inhibited by the set output signal from the latch associated with Device 1.
If Device 2 were the highest priority device requesting service, its associated latch would prime the NOR gate 17a with the reset output signal and, because the latch associated with Device 1 is not set, the NOR gate 17a would be activated, producing an output signal that primes an AND gate 18b. The lower order NOR gates are inhibited by the set output signal from the latch associated with Device 2 via the OR gates 16b, 16a, and so on. When any one of the OR gates 16 is activated, the OR gate 15 turns on and produces an interrupt request signal.
The output gates such as the AND gates 18a and 18b furnish an enabling signal to an address output device such as the switch device 19. The switch device 19 includes a plurality of transmission gates 191-193 which couple the output signal from an associated switch to a corresponding line of the data bus 110. The switches can be set to couple a logical one or zero to the corresponding data bus line when the transmission gates 191-193 are activated by the output signal from the AND gate 18a.
The interrupt cycle signal (INT CYCLE) generated in the microprocessor in response to the interrupt request (INT REQ) primes an AND gate 112 which is enabled by the TPA signal occurring during that cycle. The output signal from the AND gate 112 clocks a D-type flip-flop 116 to the set condition because the data terminal is coupled to a logical one signal (+V). The set output signal from the flip-flop 116 supplies the DMA-IN request signal. (The COSMAC architecture is arranged so that a DMA cycle is given priority after an interrupt cycle.) Following the interrupt cycle, the DMA CYCLE signal enables one of the output AND gates 18a, 18b or so on, which is primed by the priority signal. This activated AND gate turns on the associated transmission gates to couple a preset eight-bit address onto the data bus 110. The DMA CYCLE signal primes an AND gate 117 which, during the cycle's TPA signal, resets the flip-flop 116 to remove the DMA-IN request signal. The DMA CYCLE signal also primes a group of AND gates 10a-10d. The AND gates 10a-10d are also primed by the activating signal from the highest priority device. The TPB signal occurring during the DMA cycle enables one of the AND gates 10a-10d to reset the latch which initiated the previous interrupt cycle. In the example given, NOR gate 17a was enabled, priming AND gate 10b so that this AND gate becomes enabled and resets the latch 11 of Device 2. If any of the other lower order latches 11 are set, they remain set because the NOR gates 17 associated therewith are disabled. In this case, an interrupt request signal (INT REQ) is maintained which repeats the above cycle of operations (after the interrupt program is completed for the higher priority device).
The above sequence of operations continues until all of the flip-flops 11 are reset. This primes the AND gates 12 to receive the next service request signals. (The AND gates 12 prevent a higher priority latch from being reset if a higher priority latch were set just as the reset signal occurred for a lower priority device. This would cause the lower priority device to generate another interrupt because its associated latch was not reset and the higher priority device would not be serviced. Alternatively, the AND gates 12 could be inhibited during the DMA cycle which would permit high priority service requests to set their associated latches before low priority devices were serviced.)
Between each interrupt cycle and its following DMA cycle, the appropriate subroutine is executed. When an interrupt is activated, the processor inhibits further interrupts until the associated subroutine is completed.
FIG. 2 is an illustration of the logic used to modify the circuit of FIG. 1 when a two-byte branch address is to be stored. Two switch devices are required for each interrupting device and two successive DMA cycles must be generated. The flip-flop 116 and the AND gate 112 operate in the circuit of FIG. 2 in the same manner as in the circuit of FIG. 1. During a TPA pulse of the first of a pair of DMA cycles, an AND gate 21 is enabled, which sets an SR-type flip-flop 22. The set output signal from the flip-flop 22 and the reset output signal from a flip-flop 24 enable an AND gate 210, whose output signal primes one (28a) of a pair of AND gates 28a and 28b. The other input signal to the AND gates 28a and 28b is the priority decoder output signal from the high priority latch or one of the NOR gates 17a-17c. The AND gate 28a, when enabled, activates a first address word switch device 29a, which places the first byte of the two-byte address on the data bus. At the following TPB signal, the flip-flop 24 is set via an AND gate 23. The TPA signal which occurs during the second of the pair of DMA cycles enables an AND gate 212 which is primed by the set output signal from the flip-flop 24 to reset the flip flop 22. The reset output signal from the flip-flop 24 inhibits the AND gate 21 so that no set signal is applied to the flip-flop 22 during the second DMA cycle. The set output signal from the flip-flop 24 and the reset output signal from the flip-flop 22 enable an AND gate 211 whose output signal primes the second AND gate 28b of the activated interrupting device to enable the associated second switching device 29b which places on the data bus the second byte of the two-byte address.
The TPB signal occurring during the second DMA signal activates an AND gate 213 which resets the flip-flop 24. The AND gate 213 also applies an enabling signal to the reset AND gates 20d of the associated latches. The circuit of FIG. 2 produces two successive DMA-IN cycles which store two successive bytes representing the address to which the interrupt routine is to transfer. In all other respects, the circuit of FIG. 1 operates as described above.
FIG. 3 shows an alternative switch device used in place of the switch device 19. A ROM 31 is provided to couple data words selected by the status signals from the interrupting device when enabled by the output signal from the AND gate 18a as described above. This subroutine address selected from the ROM 31 and applied to the data bus depends on the reason (status) for the interrupt and provides a hardware resolution which eliminates the need for a separate resolution subroutine.
The system of the invention described above can be adopted for use in other microprocessors or computers by those of ordinary skill in the art. Those microprocessors not provided with DMA capabilities can be modified by the addition of logic networks to provide the interrupt and DMA features. (See, for example, "Increase Microprocessor Efficiency," D. C. Wyland, Electronic Design 23, Nov. 8, 1975, pp. 70-75 or "Speed Microprocessor Responses," E. Fischer, IBID, pp. 78-83.) Timing pulses like TPA and TPB can be decoded from the clock if the associated processor does not generate them.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US3665415 *||Apr 29, 1970||May 23, 1972||Honeywell Inf Systems||Data processing system with program interrupt priority apparatus utilizing working store for multiplexing interrupt requests|
|US3766526 *||Oct 10, 1972||Oct 16, 1973||Atomic Energy Commission||Multi-microprogrammed input-output processor|
|US3800287 *||Jun 27, 1972||Mar 26, 1974||Honeywell Inf Systems||Data processing system having automatic interrupt identification technique|
|US3833888 *||Feb 5, 1973||Sep 3, 1974||Honeywell Inf Systems||General purpose digital processor for terminal devices|
|US3905025 *||Apr 16, 1974||Sep 9, 1975||Ibm||Data acquisition and control system including dynamic interrupt capability|
|US3943495 *||Dec 26, 1973||Mar 9, 1976||Xerox Corporation||Microprocessor with immediate and indirect addressing|
|US3944985 *||Oct 19, 1973||Mar 16, 1976||Texas Instruments Incorporated||Workspace addressing system|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US4200912 *||Jul 31, 1978||Apr 29, 1980||Motorola, Inc.||Processor interrupt system|
|US4237535 *||Apr 11, 1979||Dec 2, 1980||Sperry Rand Corporation||Apparatus and method for receiving and servicing request signals from peripheral devices in a data processing system|
|US4240140 *||Dec 26, 1978||Dec 16, 1980||Honeywell Information Systems Inc.||CRT display terminal priority interrupt apparatus for generating vectored addresses|
|US4271466 *||Nov 21, 1978||Jun 2, 1981||Panafacom Limited||Direct memory access control system with byte/word control of data bus|
|US4310880 *||Sep 10, 1979||Jan 12, 1982||Nixdorf Computer Corporation||High-speed synchronous computer using pipelined registers and a two-level fixed priority circuit|
|US4315314 *||Dec 19, 1979||Feb 9, 1982||Rca Corporation||Priority vectored interrupt having means to supply branch address directly|
|US4403282 *||Apr 29, 1980||Sep 6, 1983||Data General Corporation||Data processing system using a high speed data channel for providing direct memory access for block data transfers|
|US4404650 *||Jan 27, 1981||Sep 13, 1983||Siemens Aktiengesellschaft||Method and circuit arrangement for transmitting binary signals between peripheral units which are connected to one another via a central bus line system|
|US4523277 *||Sep 30, 1982||Jun 11, 1985||Ncr Corporation||Priority interrupt system for microcomputer|
|US4626987 *||Aug 29, 1983||Dec 2, 1986||U.S. Philips Corporation||Method of and circuit arrangement for supplying interrupt request signals|
|US4627018 *||Sep 8, 1983||Dec 2, 1986||Sperry Corporation||Priority requestor accelerator|
|US4631670 *||Jul 11, 1984||Dec 23, 1986||Ibm Corporation||Interrupt level sharing|
|US4736336 *||Dec 23, 1985||Apr 5, 1988||Bull, S.A.||Asynchronous demand selector with multi-tape delay line|
|US4807117 *||Jul 19, 1984||Feb 21, 1989||Nec Corporation||Interruption control apparatus|
|US4901234 *||Mar 27, 1987||Feb 13, 1990||International Business Machines Corporation||Computer system having programmable DMA control|
|US4908749 *||Nov 15, 1985||Mar 13, 1990||Data General Corporation||System for controlling access to computer bus having address phase and data phase by prolonging the generation of request signal|
|US4961067 *||Jul 28, 1986||Oct 2, 1990||Motorola, Inc.||Pattern driven interrupt in a digital data processor|
|US4964034 *||Aug 25, 1988||Oct 16, 1990||Raytheon Company||Synchronized processing system with bus arbiter which samples and stores bus request signals and synchronizes bus grant signals according to clock signals|
|US5063498 *||Jul 20, 1989||Nov 5, 1991||Kabushiki Kaisha Toshiba||Data processing device with direct memory access function processed as an micro-code vectored interrupt|
|US5179696 *||Jun 24, 1991||Jan 12, 1993||Nec Corporation||Generator detecting internal and external ready signals for generating a bus cycle end signal for microprocessor debugging operation|
|US5241661 *||Jul 10, 1992||Aug 31, 1993||International Business Machines Corporation||DMA access arbitration device in which CPU can arbitrate on behalf of attachment having no arbiter|
|US5276818 *||Apr 20, 1990||Jan 4, 1994||Hitachi, Ltd.||Bus system for information processing system and method of controlling the same|
|US5280618 *||Feb 21, 1990||Jan 18, 1994||Motorola, Inc.||Interrupt test circuit for microprocessor system|
|US5317696 *||Jun 26, 1991||May 31, 1994||International Business Machines Corporation||Bus arbitration scheme|
|US5692136 *||Jul 24, 1996||Nov 25, 1997||Nec Corporation||Multi-processor system including priority arbitrator for arbitrating request issued from processors|
|US6189049 *||Aug 10, 1998||Feb 13, 2001||Micron Technology||Method for operating processor with internal register for peripheral status|
|US6219720||Aug 10, 1998||Apr 17, 2001||Micron Technology, Inc.||Core logic unit with internal register for peripheral status|
|US6233627||Aug 10, 1998||May 15, 2001||Micron Technology, Inc.||Processor with internal register for peripheral status|
|US6338098||Jan 26, 2001||Jan 8, 2002||Micron Technology, Inc.||Processor with internal register for peripheral status|
|US6374320||Aug 10, 1998||Apr 16, 2002||Micron Technology, Inc||Method for operating core logic unit with internal register for peripheral status|
|US6393507||Jan 17, 2001||May 21, 2002||Micron Technology, Inc.||Computer system including core logic unit with internal register for peripheral status|
|US7617346 *||Feb 27, 2007||Nov 10, 2009||Integrated Device Technology, Inc.||Rapid input/output doorbell coalescing to minimize CPU utilization and reduce system interrupt latency|
|US7870313||Feb 27, 2007||Jan 11, 2011||Integrated Device Technology, Inc.||Method and structure to support system resource access of a serial device implementating a lite-weight protocol|
|US8094677||Feb 27, 2007||Jan 10, 2012||Integrated Device Technology, Inc.||Multi-bus structure for optimizing system performance of a serial buffer|
|US8516163||Feb 27, 2007||Aug 20, 2013||Integrated Device Technology, Inc.||Hardware-based concurrent direct memory access (DMA) engines on serial rapid input/output SRIO interface|
|US20040046974 *||Sep 12, 2003||Mar 11, 2004||Canon Kabushiki Kaisha||Image processing device and image processing method|
|US20080205422 *||Feb 27, 2007||Aug 28, 2008||Integrated Device Technology, Inc.||Method And Structure To Support System Resource Access Of A Serial Device Implementing A Lite-Weight Protocol|
|US20080205438 *||Feb 27, 2007||Aug 28, 2008||Integrated Device Technology, Inc.||Multi-Bus Structure For Optimizing System Performance Of a Serial Buffer|
|US20080209084 *||Feb 27, 2007||Aug 28, 2008||Integrated Device Technology, Inc.||Hardware-Based Concurrent Direct Memory Access (DMA) Engines On Serial Rapid Input/Output SRIO Interface|
|US20080209089 *||Feb 27, 2007||Aug 28, 2008||Integrated Device Technology, Inc.||Packet-Based Parallel Interface Protocol For A Serial Buffer Having A Parallel Processor Port|
|US20080209139 *||Feb 27, 2007||Aug 28, 2008||Integrated Device Technology, Inc.||Rapid Input/Output Doorbell Coalescing To minimize CPU Utilization And Reduce System Interrupt Latency|
|DE3822324A1 *||Jul 1, 1988||Jul 20, 1989||Mitsubishi Electric Corp||Vorrichtung zum zerlegen des prioritaetswertes|
|EP0288607A1 *||Dec 15, 1987||Nov 2, 1988||International Business Machines Corporation||Computer system having a multi-channel direct memory access arbitration|
|U.S. Classification||710/28, 710/50|
|International Classification||G06F13/32, G06F13/34, G06F9/48|
|Cooperative Classification||G06F13/34, G06F9/4812|
|European Classification||G06F9/48C2, G06F13/34|
|Jul 13, 1994||AS||Assignment|
Owner name: MARTIN MARIETTA CORPORATION, MARYLAND
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:GENERAL ELECTRIC COMPANY;REEL/FRAME:007046/0736
Effective date: 19940322
|Jul 14, 1997||AS||Assignment|
Owner name: LOCKHEED MARTIN CORPORATION, MARYLAND
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:MARTIN MARIETTA CORPORATION;REEL/FRAME:008628/0518
Effective date: 19960128